What does SEO for a new website need before a single page is built?
The things that are expensive to change later. A new site has no history, so there's nothing to protect yet, and that's exactly why the early calls matter. Every URL, template and setting chosen now is something Google will learn, and changing it after launch means redirects, recrawls and a wait. The plan I write covers what decides whether Google can find and index every page, the same ground as my WordPress crawlability and indexing services.
Most advice on how to do SEO for a new website starts on launch day, which is late. The developer owns the build and the design, and I don't touch either. My part is the search side: which page answers which search, what the URLs look like, what each title says, and what has to be true before the domain goes live.
| Decision | What I settle | Why it's hard to change later |
|---|---|---|
| Page map | One search per page, and no two pages on the same one | Merging or splitting pages after launch means redirects and a wait while Google relearns them |
| URL structure | Short readable words, hyphens between them, a folder only where it groups real topics | Every changed URL needs a 301, and Google has to recrawl both addresses |
| Templates | Where the title, H1, canonical and schema come from on each page type | A template that prints the wrong canonical does it on every page of that type at once |
| Domain version | https, and www or not, chosen once | Switching later is a site move, with a redirect for every URL |
Google's URL guidance backs the second row: it prefers readable words to long ID numbers, and hyphens to underscores. On WordPress that means the permalink setting is picked before the first page is published. If the site's replacing an old one, it isn't really a launch. It's a move, and the old URLs need a one to one redirect map first.
How do I choose one search per page before the design starts?
With enough keyword research to give every planned page one search, and no more. A new site has no Search Console data of its own, so I start from the words customers use: the services the business sells, the towns it works in, and the questions it gets on the phone. Then I check each phrase in Google itself, looking at the suggestions, the People also ask box and the kind of page that ranks. If the results are all guides, a sales page won't fit there, and the plan says so.
Each search then gets exactly one URL. Two pages built for the same search compete with each other, so where two phrases mean the same thing to Google, they share a page.
| Search the page answers | Page | URL |
|---|---|---|
| plumber red deer | Home | / |
| emergency plumber red deer | Emergency plumbing | /emergency-plumbing/ |
| water heater replacement red deer | Water heaters | /water-heater-replacement/ |
| how much does a water heater cost to replace | A guide that answers the question | /blog/water-heater-replacement-cost/ |
That table is what the designer builds the menu from, so the page list is settled before anyone lays out a screen.
What goes on the new website SEO checklist for launch day?
This is the list I check in the page source on the live domain, the day the site goes public. Staging comes first, but nothing's signed off until it's true on the real address.
- i.
Search engines aren't discouraged
WordPress has a Reading setting called Discourage search engines from indexing this site. When it's ticked, WordPress prints a noindex, nofollow robots tag on every page. It's often ticked during the build.
- ii.
Robots.txt lets Google in
A staging robots.txt with Disallow: / can travel to the live server. I read the live file and make sure it blocks nothing meant to be found, and names the sitemap.
- iii.
Every page has its own title and description
Google's starter guide says a good title is unique to the page, clear and concise. A template that repeats the site name as every title fails that on day one.
- iv.
One H1 that names the service and the place
Not a slogan. The H1 uses the searcher's words and agrees with the title.
- v.
Canonicals name the live address
Each page's canonical points at itself, on https and the chosen www version, never at the staging domain.
- vi.
Structured data matches the business
Markup describes the business as it really is, and nothing is marked up that the page doesn't show.
- vii.
Images carry descriptive alt text
Google's starter guide calls alt text a short but descriptive piece of text that explains how the image relates to the page. A new site's photos get it before launch, not as a clean-up job later.
- viii.
Every page is linked from somewhere
Each page is reachable from the menu or another page, and the sitemap lists only live pages that return a 200.
A noindex, a Disallow or a canonical naming the staging address all look fine in a browser. That's why I read them in the source of the live pages, not in the dashboard.
Rather have me do this? One person, whole job.
Book a call on UpworkHow do I submit a sitemap to Google and get the first crawl?
Google usually finds a new site on its own. I still submit, because it hands Google the full list of pages and gives me a report I can watch from the first day.
- Step 01
Verify a Domain property
In Search Console I add the domain with a DNS record. A Domain property covers every subdomain and both http and https, and the record stays in place afterwards.
- Step 02
Open the sitemap on the live domain
I check it lists live https URLs only: no staging addresses, no noindexed pages, no redirects.
- Step 03
Submit it in the Sitemaps report
The status that comes back is Success, Couldn't fetch or a count of errors, and anything other than Success gets fixed that day.
- Step 04
Name it in robots.txt
A Sitemap: line with the full URL goes into robots.txt, so any crawler that reads the file finds the list.
- Step 05
Inspect the pages that matter most
I run the homepage and main service pages through URL Inspection and request indexing for that handful. Google says a request doesn't guarantee indexing, so the sitemap does the rest.
Google's own wording is that a sitemap is merely a hint. So I don't read Success as indexed. It means Google fetched and read the file, and what's actually in the index shows up in the Pages report over the following weeks.
What should SEO for a new website look like in the first 90 days?
Quiet, steady work, judged on the right numbers. Google's starter guide says some changes take effect in a few hours and others take several months, and that you likely want to wait a few weeks to assess your work. So I won't promise a date, and the plan for how long SEO takes to work is written around what can be measured.
- i.
Weeks one to four: impressions and click rate
Traffic's too thin to judge, so I read whether Google shows the right page for the right search, and whether people click. One fireplace business that launched on 9 August 2026 shows the pattern: in the 28 days to 12 September its homepage ranked at 4.0 with a 10.46% click rate. Its case study carries a caveat that applies here too: a month is too short to call a trend.
- ii.
Keep the page map as the site grows
Every new page gets checked against the map before it's published, so it takes a search no other page already answers.
- iii.
One new page per real buyer question
Not a fixed quota of blog posts. When customers keep asking something the site doesn't answer, that question becomes a page.
- iv.
The first links come from people who know you
Suppliers, trade associations, local partners and the businesses you already work with. Google's guide lists social media, community engagement, advertising and word of mouth as ways to promote a new site, and says word of mouth takes time.
- v.
Around day 90: judge the trend, not the week
By then there's usually enough data to see which pages are climbing and which have stalled, and a stalled page gets its title, its copy and its links looked at in that order.
| Signal | What it tells me | What I do |
|---|---|---|
| Impressions by page | Whether Google's showing the page at all, and how often | A page with none after several weeks gets its indexing status checked first |
| Click rate by page | Whether the title and description match the search | A low rate at a decent position means rewriting the title, not the page |
| Queries by page | Which searches Google thinks the page answers | A page showing for the wrong search usually has the wrong title or H1 |
| Pages report | Which URLs are indexed, and why the rest aren't | Anything that should be indexed and isn't gets its status read one URL at a time |
Launching soon? I can own the search side from the plan onwards: the page map, the checks on the live domain and the first 90 days of Search Console, all remotely, while your developer keeps the build.
Book a call on UpworkWhich mistakes are hardest to undo after launch?
The ones that change addresses. A title can be rewritten in a minute and Google picks it up on the next crawl. A URL can't be changed without a redirect, and a site that renames its pages three months in has to wait while Google works out that old and new addresses are the same page. Next comes the page map, because two pages built for one search have to be merged later, and then the domain version, because switching from www to the bare domain turns a settled site into a site move.
The launch day mistakes, a noindex left on from staging or a Disallow in robots.txt, are the most common and the cheapest to fix. Cheap isn't free, though: every day they're live, Google's been told to stay away.
